engleski

Revival of filtering based SLAM? Exactly sparse delayed state filter on Lie groups

Simultaneous Localization And Mapping (SLAM) is a core element of every modern mobile autonomous robot. The underlying engine of a SLAM system is its back-end, which aims at optimally estimating trajectory and the map of the robot’s environment based on sensor data abstractions. Over the past 10 years filtering based SLAM solutions gave in to graph optimization approaches, since the latter dominated in performance over a wider range of applications. In this paper we propose a novel SLAM back-end based on the exactly sparse delayed state filter (ESDSF) and the extended Kalman filter on Lie groups (LG-EKF). Using LG-EKF directly would yield to same limitations as the early EKF-based SLAM approaches ; therefore, we derive the ESDSF on Lie groups, which we dub LG-ESDSF. The proposed filter retains all the good characteristics of the classic ESDSF, but also respects the state space geometry by employing filtering equations directly on Lie groups. We have compared our SLAM system with two current state-of-the-art SLAM solutions, namely ORB-SLAM and LSD-SLAM, on the KITTI vision benchmark suite. Results showed that the proposed SLAM based on the LG-ESDSF back- end can match and even outperform methods based on graph optimization techniques.

SLAM, ESDSF, Lie groups

nije evidentirano